When addicts perceive conditioned stimuli (CSs) associated with drugs, they often find themselves engaging in drug-seeking behavior, even after a period of successful abstinence. Understanding how CSs promote drug- and reward-seeking behavior is thus of primary importance to understanding addiction and relapse. In this proposal, we focus on conditioned approach behavior: the conditioned locomotor response to the CS that often brings the subject closer to the predicted reward. These responses can be flexible in that the specific approach actions required to reach the target location can vary across encounters with the CS. We use a rat model of flexible conditioned approach to investigate the neural circuitry underlying this behavior. An important element of this circuitry is the nucleus accumbens (NAc), a brain region that contributes significantly to addictive behavior. Part of this contribution is likely due to the critical role it plays in flexible approach responses: these absolutely depend on the NAc and the dopamine projection it receives from the ventral tegmental area. However, it is not yet understood how NAc neurons dopamine-dependently facilitate flexible approach. One possibility (Target Selection) is that some NAc neurons selects among different possible targets to approach based on the reward predicted by the CS and the target location. Other NAc neurons could serve a different function (Approach Gating) of activating downstream circuits that determine the target and the approach actions, without themselves contributing to target selection. To understand conditioned approach behavior, it is essential to know whether NAc neurons serve only one of these functions (and if so, which one) or whether they participate in both of them. To test the Target Selection and Approach Gating hypotheses, we use multiple CSs, multiple targets to approach, and/or multiple outcomes to determine whether the CS-evoked firing responses of NAc neurons in behaving rats encode information about the approach target, the outcome predicted by the CS, or both. Next, we determine how the dopamine input to the NAc influences this encoding, using a powerful new technique developed in our lab for applying pharmacological compounds (dopamine antagonists and agonists) to the neurons we record from in behaving animals. In particular, we test the long-standing (but not yet directly tested) hypothesis that dopamine contributes the reward-predictive component of NAc neurons' encoding of CSs. Our technique allows us to establish different potential contributions of D1 and D2 classes of dopamine receptors to behaviorally-relevant NAc neuronal firing. Thus, these experiments will reveal specific circuit mechanisms whereby NAc neurons promote approach responses to CSs. By doing so, they will enhance our understanding of the control over addicts' drug-seeking behavior by drug-associated cues.